Evolutionary History, Habitat Disturbance Regimes, and ...

Copyright © 2009 by the author(s). Published here under license by the Resilience Alliance. Waples, R. S., T. Beechie, and G. R. Pess 2009. Evolutionary history, habitat disturbance regimes, and anthropogenic changes: What do these mean for resilience of Pacific salmon populations? Ecology and Society 14(1): 3. [online] URL: http://www.ecologyandsociety.org/vol14/iss1/art3/

Synthesis, part of a Special Feature on Pathways to Resilient Salmon Ecosystems

Evolutionary History, Habitat Disturbance Regimes, and Anthropogenic Changes: What Do These Mean for Resilience of Pacific Salmon Populations? Robin S. Waples 1, Tim Beechie 2, and George R. Pess 2

ABSTRACT. Because resilience of a biological system is a product of its evolutionary history, the historical template that describes the relationships between species and their dynamic habitats is an important point of reference. Habitats used by Pacific salmon have been quite variable throughout their evolutionary history, and these habitats can be characterized by four key attributes of disturbance regimes: frequency, magnitude, duration, and predictability. Over the past two centuries, major anthropogenic changes to salmon ecosystems have dramatically altered disturbance regimes that the species experience. To the extent that these disturbance regimes assume characteristics outside the range of the historical template, resilience of salmon populations might be compromised. We discuss anthropogenic changes that are particularly likely to compromise resilience of Pacific salmon and management actions that could help bring the current patterns of disturbance regimes more in line with the historical template. Key Words: climate change; duration; frequency; historical template; magnitude; Pacific Northwest; Oncorhynchus; Pacific salmon; predictability.

INTRODUCTION The concept of resilience means different things in different contexts. In the field of communications technology, resilience is the ability of a network to keep functioning and provide services in spite of disturbances to normal operations. In this context, resilience is more or less synonymous with robustness. To physicists, resilience takes a meaning more like elasticity and describes the speed or fidelity with which a material returns to its original state after it has been deformed. Ecologists have often used resilience to refer to a similar phenomenon: the ability of a biological system to return to equilibrium after a perturbation. However, Holling (1973) argued that stability is a better term for this quality, and that the concept of ecological resilience more properly refers to the ability of a system to absorb change and still maintain its basic system of relationships without flipping into a different configuration. In this view, resilient biological systems might be inherently unstable but nevertheless persistent. Holling (1973:18) also noted that “the balance between resilience and 1

stability is clearly a product of the evolutionary history of these systems in the face of the range of random fluctuations they have experienced.” This quote emphasizes the important point that ecological resilience incorporates both forward- and backward-looking components: we typically think of resilience as a property that describes future behavior of a biological system, but that behavior is the consequence of attributes of the system that have been forged over time in the crucible of past evolutionary events. In this paper, we consider how the evolutionary history of Pacific salmon (Oncorhynchus spp.) and the dynamic ecosystems they inhabit have conferred considerable resilience on this important natural resource. We include in the term “Pacific salmon” not only the five species traditionally recognized from North America (pink salmon, O. gorbuscha; chum salmon, O. keta; coho salmon, O. kisutch; sockeye salmon, O. nerka; Chinook salmon, O. tshawytscha) but also steelhead, the anadromous form of rainbow trout, O. mykiss. Pacific salmon have evolved in an environment that is both

NOAA Fisheries Northwest Fisheries Science Center, 2NOAA Fisheries, Northwest Fisheries Science Center Environmental Conservation Division

Ecology and Society 14(1): 3 http://www.ecologyandsociety.org/vol14/iss1/art3/

physically and climatically dynamic, with the major extant lineages developing during the last ice age, which occurred ~400,000 yr to 16,000 yr ago, and current metapopulation structure developing during a period of rapid landscape evolution ~16,000 yr ago to present (Waples et al. 2008). Over the past two centuries, however, rapid climate change and major anthropogenic modifications to salmon ecosystems have dramatically altered disturbance regimes that salmon experience. To the extent that these disturbance regimes assume characteristics outside the range of the historical template the species evolved under, resilience of salmon populations might be compromised. We discuss anthropogenic changes that are particularly likely to compromise resilience of Pacific salmon and management actions that could help bring the current patterns of disturbance regimes more in line with the historical template. Resilience can be assessed at several spatial and temporal scales. For example, a variety of factors can affect salmon viability and make local spawning populations either more or less resilient to environmental fluctuations or anthropogenic changes (McElhany et al. 2000). And the degree to which the responses of different populations to perturbations are uncorrelated can confer resilience on larger geographic scales (Good et al. 2008). Furthermore, this latter type of resilience has a deeper temporal dimension that can extend well beyond the expected persistence time of any individual population. THE HISTORICAL TEMPLATE Waples et al. (2008) recently reviewed major features in the evolution of Pacific salmon and their habitats. Key points that emerged from this joint analysis include the following: 1. Speciation within Oncorhynchus was complete by the late Miocene, leaving several million years for evolution within each of the salmon species. However, none of the extant, intraspecific lineages are anywhere near that old. Several lineages probably date from the Pleistocene. 2. A great deal of diversity has developed during the Holocene, through a combination of recolonization of deglaciated habitats, e.g., Puget Sound and most of British Columbia, and in situ evolution.

3. Major, but rare, habitat upheavals such as late-Pleistocene megafloods and volcanism have had long-lasting impacts on salmon evolution. 4. In many areas, postglacial rebound and associated habitat changes stabilized by about 5000 yr ago, leaving ample time for contemporary salmon populations to have reached a dynamic equilibrium between genetic drift, migration, and local adaptation. 5. The spatial and temporal scales on which disturbance regimes operate have provided the framework within which the processes shaping Pacific salmon evolution have operated. 6. Recent anthropogenic changes have altered characteristics of disturbance regimes and created novel evolutionary pressures for Pacific salmon. What features of the historical template are important for resilience of Pacific salmon? Here, we consider both the physical and biological aspects of this problem. Dynamic attributes of salmon habitats Salmon habitats are dynamic in many ways, including annual movement because of channel migration, shifting locations of wood accumulations that form pools, and changes in structure and function of riparian forests as floodplain surfaces erode and form anew elsewhere; in addition, shortterm perturbations of habitats occur during floods or low flow periods. These processes create a dynamic mosaic of habitats to which salmon are adapted, and disturbance regimes are fundamental in shaping those environments (Stanford et al. 1996, Independent Scientific Group 1999, Bisson et al. in press). Disturbance regimes can be characterized by four main attributes (Lytle and Poff 2004): (1) how often the event occurs (frequency); (2) spatial extent and severity (magnitude); (3) the length of the disturbance; press and pulse (duration); and (4) novel to routine (predictability). These attributes can be applied to each of the five main drivers of habitat condition and use by salmon: sediment supply regime, hydrologic regime, thermal regime, riparian vegetation, and connectivity (Table 1), as well as to river channels and habitats themselves.


However, the characterization of each disturbance regime is scale dependent, because a disturbance at a small scale, such as a landslide temporarily killing off a small spawning population in a single tributary, is simply part of the normal pattern at a larger scale. For example, landslides occur every year at a few locations within a large river basin. Because the attributes of disturbance regimes are intertwined, it is difficult to separate them cleanly and assess how salmon are resilient to the various components. Moreover, each of the five main drivers has multiple disturbance attributes, as well as multiple directions of human impact on the regime. For example, hydrologic regime encompasses both peak and low flows, each of which influences salmon population performance, and human impacts can either reduce or exaggerate both attributes. Dam operations might dampen peak flows and increase low flows, whereas urbanization or climate change might increase peak flows and decrease low flows. Hence, it is difficult to summarize all the characteristics of natural disturbance regimes in a simple way, and to assess how human activities alter each aspect of disturbance regime. Here we describe some key aspects of disturbance regimes that salmon experience to set the stage for assessing their responses to dynamic environments. Hydrologic and sediment supply regimes that occurred prior to European settlement were typified on the one hand by high-magnitude, low-frequency events such as floods and landslides, and on the other hand by infrequent periods of low flow or sediment supply (Reeves et al. 1995). Floods occur annually throughout the Pacific Northwest, and most extreme events tend to occur either during fall and winter rain storms combined with snowmelt (Sumioka et al. 1998), or during localized stormrelated flood events in spring and summer. Low flows for most salmon populations occur in late summer and early fall, and in some regions long reaches of river are dry for weeks to months during that period. High-intensity rainfall events are the primary drivers of erosion, both in coastal areas where landslides dominate and in the interior where surface erosion dominates (Beechie et al. 2003). Erosion events can reduce salmon survival in a local area for up to several years, but most are too small to significantly alter channel morphology and habitat structure. By contrast, long periods of low erosion intensity can lead to sediment-poor stream

conditions and lack of spawning and rearing habitat. Although this is not a common occurrence within the range of Pacific salmon, such episodes of low sediment supply can persist for decades to centuries until fires and storms conspire to increase sediment delivery rates (Benda and Dunne 1997). Riparian vegetation varies from grasses and willow or brush species in semiarid regions to dense stands of large conifers in coastal rain forests. Disturbance regimes are driven by two main processes: fire in relatively small streams (Beechie et al. 2000) and channel migration and bank erosion in larger rivers (Beechie et al. 2006a, Laterell et al. 2007). Hence, wood recruitment regimes vary from minimal quantities in semiarid regions (excepting beaver use of small wood to build beaver dams; Pollock et al. 2007) to large and consistent annual inputs in which floodplains continuously erode patches each year and incorporate trees into the channel (Latterell et al. 2007). Wood recruitment is most episodic in small streams, in which bank erosion is minimal and wood is delivered in pulses as a result of fires. Although this paper will focus on freshwater habitats, it is worth noting that ecological and evolutionary processes in salmon populations are also shaped by natural disturbance processes in estuaries and near-shore marine areas. These habitats are also dynamic on both short and long time scales as a result of geological processes (e.g., subduction-zone earthquakes), and oceanographic processes (coastal upwelling, ENSO oscillations, etc.). In contrast to discrete ecosystems such as lakes, therefore, the ecosystems that Pacific salmon depend upon are open systems with fluid boundaries (Bottom et al. 2009), a reality that presents special challenges for assessing and managing resilience. Biological attributes of salmon populations Collectively, these patterns of dynamic change to their habitats represent the evolutionary milieu of Pacific salmon. In this section, we outline some of the biological attributes of salmon populations that contribute to their resilience in the face of these patterns of change. We refer the reader to Groot and Margolis (1991), Hendry and Stearns (2004), and Quinn (2005) for detailed treatments of salmon biology, and to Healey (2009) for additional discussion of how these factors affect resilience.


Table 1. Descriptions of typical natural and altered disturbance regimes.

Natural

Altered

Connectivity within natural range Magnitude

Landslide dams may block migration to very small, e. Dams and culverts block small and g., a single tributary or large areas, e.g., most of the large areas, and thousands of Fraser River blockages are in place simultaneously.

Frequency

Migration blockages are rare, < once every thousand years in most locations

Not frequent, i.e., each blockage goes in once, but many areas blocked simultaneously

Duration

Most landslide dams are temporary, lasting hours to days.

Tens to hundreds of years to date

Predictability

Locations are not very predictable.

Once in place the blockage is “predictable”.

Magnitude

Magnitude of sediment supply varies spatially and temporally, driven by precipitation intensity, duration, and extent.

Sediment supply generally increases due to land uses such as forestry, grazing, or cultivation.

Frequency

Sediment supply is episodic. Some sediment enters a river system each year, but location, amount, timing, type, size, etc. vary from year to year.

Land uses such as forestry tend to increase the frequency of sediment inputs to rivers.

Duration

Duration of sediment supply generally mirrors storm durations.

Little change from land use

Predictability

Stochastic

Stochastic, but more frequent

Sediment supply

Hydrologic regime Magnitude

Spatial extent of storms and associated floods can be Typically reduced due to dams as small as a single watershed, or as large as an entire absorbing the peak flows region, e.g., the 1964 storm, in which record floods occurred from northern California to southern Washington.

Frequency

Large floods or extreme low flows typically occur several times in a given year.

Peak and low flows are typically reduced at the seasonal and annual time scale, whereas diurnal range may be increased at the daily time scale.

Duration

Ranges between hours, days, and weeks

Rivers regulated by dams typically increase the duration of peak flows.

Predictability

Highly predictable on an annual scale, e.g., whether they will occur; less predictable at the monthly, weekly, or daily scale

Greater predictability due to rivers being regulated by dams

Temperature extremes rarely exceed tolerances of most salmon populations.

Temperature extremes commonly exceed tolerances of many salmon populations.

Thermal regime Magnitude

(con'd)


Frequency

No exceedence in most years at most locations

Several times a year in most locations

Duration

Days to weeks

Days to weeks

Predictability

Predictable

Predictable

Magnitude

Small streams: Fires periodically kill riparian forest patches. Large rivers: River erosion periodically removes patches of floodplain forest.

Fires and erosion rarely kill riparian forest patches, but logging and land conversion kill a greater expanse of riparian forests.

Frequency

Small streams: fire reset vegetation at return intervals of 50–400 yr Large rivers: erosion reset forests at intervals of 1 to ~100s of yr

Logging: return interval of < 50 yr in most cases, riparian protections recently enacted in some areas Land conversion: generally kills forests once

Duration

Forests begin regeneration soon after disturbance.

Logging: forests begin regeneration soon after disturbance. Land conversion: often no regeneration, i.e., duration of decades to centuries

Predictability

Fire: unpredictable Erosion: predictable

Predictable

Magnitude

Millions of salmon returning

Few to no salmon returning in many areas

Frequency

Annual

Annual

Duration

Salmon returning in most months of the year

Salmon returning in most months, but shorter spawning runs for many populations

Predictability

Predictable

Predictable

Riparian vegetation

Nutrient regime

Life history diversity Pacific salmon exhibit a rich variety of life history traits, both within and among populations. Some of the traits that are most important for resilience are summarized in Table 2. When juvenile salmon migrate from fresh water to the sea they are known as smolts, and this can occur anywhere from a few days to two or more years after they emerge from the gravel in which the female salmon deposits her eggs. In some species, a fraction of individuals returns to fresh water to spawn the same year they emigrate as smolts, but most spend one or more winters at sea before initiating their spawning migration. Together, these two traits define the age

structure of a salmon population. Two species, O. nerka and O. mykiss, have forms that spend their entire life cycles in fresh water without ever migrating to sea. This trait can be characteristic of an entire population, or it can be a polymorphism within a single population. Finally, all ‘true’ Pacific salmon die after spawning, but resident forms of O. mykiss are iteroparous, and some fraction of anadromous steelhead also spawn more than once. All else being equal, populations that can express any of several juvenile and marine life history strategies will be less strongly affected by extreme environmental events, good or bad, that affect a single year class. Likewise, if a population complex


Table 2. Occurrence of major life history traits in native Pacific salmon populations from North America. Symbols represent proportion of populations that predominantly express the indicated trait: +++ majority; ++ approximately half; + minority; - trait is absent in the species; * a minority of individuals in some populations express the trait. Modified from Waples et al. (2001).

Species

Age at smolting

Winters at sea

Anadromous

Spawning

1

1

Yes

No

Semelparous Iteroparous

Pink

All

-

-

-

All

-

All

-

All

-

Chum

All

-

-

-

-

All

All

-

All

-

Sockeye

+

+

+++

-

+

+++

+++

+

All

-

Coho

*

+++

+a

*b

All

*

All

-

All

-

Chinook

++

++a

*

*b

+

+++

All

*

All

-

Steelhead

-

+

+++

*

++

++

++

++

+++c

+

aSmolt age in these species is generally older in populations from central B.C. to the north. bOnly males express this trait. cIn most pure steelhead populations, the fraction of individuals that spawns more than once

contains populations that have different suites of life history traits, overall abundance can be buffered against environmental fluctuations that occur on a variety of temporal scales (Hilborn et al. 2003, Koski et al. 2009). Moreover, populations with diverse life histories can take advantage of habitats that cannot support a population through all its freshwater life stages. For example, coho “nomads,” (fry outmigrants) rear in estuaries for their first summer, then migrate into tributaries to overwinter. This life history strategy allows coho smolts to be produced from streams that lack suitable spawning habitat, and conservation of both estuarine and overwinter habitats is crucial to maintaining high smolt production from streams that otherwise could not support coho salmon (Koski et al. 2009).

is small.

Homing and straying

between homing and straying helps to shape the adaptive landscape for Pacific salmon. Homing fidelity is strong enough that local populations can become adapted to their particular environments (Taylor 1991). However, except for populations that become isolated above natural barriers, e.g., waterfalls created by glacial rebound, most salmon populations are part of larger metapopulations that are connected by migration. This ensures that local populations do not become too small and isolated and that each contains an appreciable fraction of the genetic variation contained by the species. Furthermore, straying provides a source of colonists that can promote recovery after major disturbances and local population declines (Leider 1989). This behavior helps salmon to persist in highly dynamic environments and to rapidly expand their range when conditions permit (Milner and Bailey 1989).

Pacific salmon are justifiably renowned for their ability to return to spawn at the precise location at which they were hatched. However, homing is not perfect, which leads to a generally low level of natural straying among populations. The interplay

Sockeye salmon exhibit both extremes on the continuum of homing vs. straying. Most sockeye populations are associated with lakes, which promotes isolation, a stronger degree of population genetic differentiation than is found in other Pacific


salmon, and highly specialized adaptations (Burgner 1991, Wood 1995). Although lake-type sockeye populations can be enormously productive (Foerster 1968), these specialized habitats are vulnerable to environmental changes. Conversely, river-type sockeye are more generalized in their habitat requirements but also relatively rare, or perhaps only poorly documented, and only weakly differentiated based on neutral genetic markers (Wood 1995, Gustafson and Winans 1999, Wood et al. 2008). Although highly specialized, most laketype sockeye populations might be evolutionary dead ends, and it might be the river-type populations, which are more likely to produce successful colonists, that confer more resilience on the species as a whole (Wood 1995, 2007, Wood et al. 2008; but see Pavey et al. 2007). Phenotypic plasticity Phenotypic plasticity is the ability of the same genotype to produce a different phenotype under different environmental conditions. Expression of most of the life history traits in Table 2, as well as many others, can be strongly affected by factors such as water temperature, food availability, growth rate, etc. Plasticity can be expressed at the level of the individual, for example, growth rate of individual fish typically varies with temperature, and integrated up to the population level. At the individual level, plasticity allows a salmon to make behavioral or physiological adjustments that allow it to thrive through a range of environmental conditions. The relationship between the expression of particular phenotypes and their associated environments is known as an individual’s norm of reaction (Hutchings 2004). If reaction norms differ among individuals within a population, the capability of the population for a plastic response to environmental change is further enhanced, without the requirement for genetic change. Rapid evolution Each species of Pacific salmon comprises a number of evolutionarily significant units (ESUs; Waples 1991, 1995), which are groups of populations that share life history/genetic/ecological traits to a substantially greater degree than do populations from different ESUs. Salmon ESUs are meant to represent units that follow largely independent evolutionary trajectories over hundreds or thousands of years. However, evidence is accumulating to show that Pacific salmon are also

capable of relatively rapid evolution, on human time scales. It seems, therefore, that within a century or so, perhaps less in some cases, genetic and phenotypic differences can be produced that are comparable to those found between Pacific salmon populations from within the same ESU (Quinn et al. 2001, Waples et al. 2004). SYNTHESIS Joint consideration of temporal and spatial scales of physical and biological changes in salmon populations and their habitats indicates that the historical template was really a shifting mosaic, with constant changes overlaid on some more robust underlying processes. On a very local scale (a single deme, or group of spawners within a population) the chances are small of a major habitat disturbance occurring during any particular year, but over long periods of time the probability of a severe event approaches unity. High productivity and variable age at maturity can help buffer demes against severe disturbances, but periodic extinction of such units is part of the evolutionary history of salmon. As the geographic scale under consideration increases to populations, metapopulations, or larger conservation units or ESUs, it becomes more likely that a severe disturbance will occur somewhere during any given time period, but the effects on the overall metapopulation or conservation unit will be modulated by normal conditions elsewhere. salmon in nearby tributaries often will survive local disturbances and can provide strays to repopulate extirpated demes and populations. Resilience of the species as a whole is thus enhanced by the existence of many demographically independent populations replicated across the landscape. Together, these physical processes and the biological attributes of salmon populations promote local adaptations, while at the same time making it unlikely that any particular locally adapted population will persist indefinitely. Therefore, if one were able to travel back in time and take a series of snapshots of habitat features and associated salmon biology, one might expect to find the following: (1) at any point in time, a large fraction of salmon populations will have acquired genetically based adaptations to local conditions; but (2) fine-scale patterns of adaptation would change over time. That is, each time period might be characterized by a general pattern of locally adapted populations, but specific habitat features


would have evolved between time periods and as a consequence the biological attributes of individual populations would evolve as well. Thus, although specific local adaptations between salmon and their habitats are not necessarily stable over long time periods, the general pattern of locally adapted populations spread across a dynamic landscape is much more robust, and it is this broader pattern that is one of the major factors that confers resilience to the system as a whole. Not all species are affected in the same way or to the same extent by these processes. A useful way to evaluate the effects of a species’ life history on resilience is by using the four criteria that characterize viable salmonid populations (VSP; McElhany et al. 2000): abundance, productivity, spatial structure, and diversity (genetics, life history, ecology). All four criteria are evaluated at the level of individual populations, and the last two criteria are also used to help guide the process of integrating the population data into an overall assessment of viability of larger conservation units and ESUs. At one extreme of the life history diversity of Pacific salmon, Pink salmon show little variation either within or between populations (Table 2). Fixed age at maturity provides no opportunity for different cohorts to fill in for catastrophic losses. In this species, resilience primarily is a product of high abundance and productivity and relatively high straying rates, which create a metapopulation structure that serves to spread the risk across multiple streams. This species migrates to sea shortly after emergence and hence avoids many problems associated with rearing in dynamic freshwater habitats. For Pink salmon, therefore, diversity is limited and is not as important a contributor to resilience as are the other criteria such as abundance and productivity. At the other end of the spectrum, Chinook salmon and steelhead support a rich diversity of life history types, both within and among populations. These species penetrate into higher elevation tributaries, in which opportunities for isolation and local adaptation are greater, and as juveniles they spend longer periods in fresh water than do Pink or chum salmon. Chinook salmon can spend up to 5 yr at sea and steelhead can spawn more than once; this wide variation in age at maturity provides ample opportunity for other year classes to compensate for a decimated cohort. Abundance of individual populations typically is lower than for Pink salmon, but the rich diversity of life history types makes it

likely that at least some will be successful under any given set of environmental conditions. For Chinook and steelhead, therefore, spatial structure and diversity play particularly important roles in promoting resilience. O. mykiss and O. nerka also have resident as well as anadromous forms, which provides an additional component of diversity and an additional layer of flexibility to respond to challenging environmental conditions. Downstream migrants from resident populations can accelerate the range and rate of colonization after disturbances events, because of positive spawning interactions with upstreammoving anadromous populations. On the Olympic Peninsula in Washington State, interactions between the two forms of O. mykiss, female steelhead and resident male rainbow trout) occur primarily at the end of the spawning season (McMillan et al. 2007). In controlled experiments, female steelhead × resident rainbow crosses can produce over 25% of the offspring that migrate to sea as smolts (Ruzycki et al. 2003, unpublished report). Other species that have resident life forms, such as O. nerka (kokanee), have shown contributions to anadromous life forms, but at a much smaller percentage (